Simultaneous localization and mapping (SLAM) is a key issue in robotic autonomous navigation systems. Traditional SLAM algorithms are usually based on assumptions of static environments, while practical applications of robots are usually complex dynamic scenarios, where traditional algorithms will fail. To address the problem that the localization and mapping accuracy of multi-robot systems decreases due to the interference of moving objects in dynamic environments, a multi-robot cooperative SLAM algorithm integrating dynamic landmark information is proposed. The algorithm constructs new inter-robot constraints by using mutual observations between robots and observations of the same dynamic landmark by different robots. Then, the odometry measurement, landmark observation, mutual observation between robots and dynamic object data association are used to formulate the collaborative factor graph optimization problem. The proposed algorithm enhances the accuracy of multi-robot system state estimation through multi-constraint coupling optimization. Finally, the algorithm is verified to be effective in dynamic scenes through simulated dynamic scenes experiments and real-world dynamic scenes experiments based on a robot platform.